Zhiqiang Wang

The vanilla cnn downsampling architecture cannot recover spatial information of a image

1

The convolutional part of the architecture act as a encoder part, it capture image's contexture information, the architecture should ensemble a decoder part (deconvolution layer or RNN layer) to recover...

[RFC] Data Augmentation of object detection

3

## 🚀 Feature Hi, Recently I found a repo about data augmentation with excellent programming, named [TensorPipe](https://github.com/kartik4949/TensorPipe). But it's implemented with tensorflow, do you mind to add this similar features...

[RFC] Support YOLOX detection model

20

### 🚀 The feature YOLO aka. You Only Look Once, which is a vibrant series of object detection models since the release of Joseph Redmon [You Only Look Once: Unified,...

Train models from scratch

11

## 🚀 Feature Support training models from scratch, this is a follow-up issue of #16. ## Motivation Test whether the trainer mechanism works. ## Pitch - [x] #67 - [x]...

### 🚀 The feature Seems that the boxes decoding computation can be fused at: https://github.com/zhiqwang/yolov5-rt-stack/blob/fb7e12039a9af3a37d1d6ea61505ca39aecf349f/yolort/relay/logits_decoder.py#L49-L60 ### Motivation, pitch Accelerated computation for `LogitsDecoder`. ### Alternatives _No response_ ### Additional context _No...

Close #60

### 🚀 The feature Do the score filter first at https://github.com/zhiqwang/yolov5-rt-stack/blob/fb7e12039a9af3a37d1d6ea61505ca39aecf349f/yolort/models/box_head.py#L348-L357 and https://github.com/zhiqwang/yolov5-rt-stack/blob/fb7e12039a9af3a37d1d6ea61505ca39aecf349f/yolort/models/box_head.py#L413-L416 ### Motivation, pitch Accelerated post-processing computation. ### Alternatives _No response_ ### Additional context _No response_

### 🚀 The feature Support exporting yolort to more inference backends. - [x] https://github.com/apache/tvm/releases/tag/v0.8.0 (TorchScript-Relay) - [x] https://github.com/NVIDIA/TensorRT/tree/main/tools/onnx-graphsurgeon (ONNX-TensorRT) - [ ] https://github.com/alibaba/TinyNeuralNetwork (TorchScript-TFLite) - [ ] https://coremltools.readme.io/docs/pytorch-conversion (TorchScript-coreML) -...

PPLNN, which is short for "PPLNN is a Primitive Library for Neural Network", is a high-performance deep-learning inference engine for efficient AI inferencing. https://github.com/openppl-public/ppl.nn _Originally posted by @xinsuinizhuan in https://github.com/zhiqwang/yolov5-rt-stack/issues/390#issuecomment-1104759564_